The number one problem I see in most newly minted data scientists is that they think their success depends on how well they know sophisticated modeling techniques, instead of how well they prepare and understand the data. Academic training usually starts with precleaned, high-quality datasets, and the students are tested on how to pick the most performant regression (i.e., math) technique. They will have intense discussions and even proofs about whether a convolutional neural network converges on a solution more quickly than a recurrent neural network or whether L1 versus L2 regression is best (these terms don’t matter in 99 percent of applications so I won’t define them here). I’ve seen Data Science projects ...
Get How to Talk to Data Scientists now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
